Visa lottery study finds H-2B caps shrink US firms without creating American jobs

A peer-reviewed study of a US government visa lottery concludes that restricting low-skill foreign labor makes American businesses smaller rather than pushing them to hire more Americans. The paper, by economists Michael A. Clemens and Ethan G. Lewis, appears in the July 2026 issue of the American Economic Journal: Applied Economics, and its findings are laid out in a research summary from the American Economic Association that recently reached the Hacker News front page.

A crashed server created a natural experiment

The H-2B visa covers temporary foreign workers in seasonal industries such as landscaping, seafood processing, forestry and hospitality. According to the AEA summary, 98% of the jobs involved require no high school education. The Immigration Act of 1990 set an annual ceiling of 66,000 visas, a figure Clemens traced, through a chain of former congressional staffers, to a subcommittee chairman who said the law's designers simply tripled the roughly 22,000 visas issued the year before and assumed the cap would never bind.

It does bind. For the second half of 2022, the AEA reports, employers requested 136,555 workers for just 33,000 available slots, and certified petitions came from all 50 states plus Washington, DC and Puerto Rico.

The randomization arrived by accident. After a surge of petitions crashed a Department of Labor server in 2019, the agency began processing employers' requests in randomized order, assigning each a letter. Firms drawing an A almost always secure the workers they asked for; firms drawing later letters generally do not. That gave the researchers something close to a controlled trial between similar firms whose main difference was luck. Clemens and Lewis surveyed 472 businesses that entered the 2021 and 2022 lotteries and registered a pre-analysis plan, fixing their hypotheses and methods before any responses arrived to guard against cherry-picking results.

Winners grew; losers did not hire Americans instead

The measured effects were substantial. Losing the lottery cut a firm's H-2B employment roughly in half. Firms that won and hired everyone they wanted saw revenue rise with an elasticity of about 0.20, meaning a doubling of H-2B employment lifted revenue by roughly a fifth. Investment in equipment, vehicles and structures was far more responsive, with an elasticity of 1.5 to 2.1.

What did not happen matters just as much. According to the AEA, employment of low-skill American workers at losing firms did not rise. Across the full sample, the effect of foreign hiring on US employment was zero to positive, and in a pre-specified subsample of rural firms it was significantly positive, with an elasticity of 0.61: just over half a percent more US workers for every one percent increase in H-2B workers.

"If you don't allow firms to hire immigrant workers, they just become smaller firms," Lewis told the AEA. "They are not replaced with US workers."

Substitution loses to scale

The economics turn on two opposing forces. At a fixed level of output, some American workers would indeed step into jobs vacated by foreign workers, the substitution effect. But firms denied workers do not hold output fixed; the AEA's illustrations include a traveling carnival that visits fewer towns and a fishing operation that covers less of the salmon season. That contraction, the scale effect, overwhelms substitution and leaves fewer jobs overall, Americans included.

Consistent with this, the authors estimate the elasticity of substitution between H-2B and US workers at roughly 0.8 to 2.2, far below the 4 to 10 typical of related studies of immigrants in low-skill work and nowhere near the perfect substitutability assumed in some influential models.

Why it matters

The 66,000 cap has gone unrevised since 1990, and this evidence suggests a marginal increase in H-2B visas would expand production, investment and profits at US firms without reducing American employment. The authors caution that their short-run figures may understate the gains, since firms facing chronic visa uncertainty tend to underinvest, and related work finds that losing firms are more likely to shut down entirely. For Clemens, the study is an argument for shared evidence in a polarized debate: "When policy is politically polarized, you need more facts," he told the AEA.